They can connect a common-name person in one place to another:
In 1840, Harry Porter "of Farmington, Fulton County, Illinois," sold property in Clarkson, Monroe County, New York, where he had lived for almost twenty years before heading west. This contemporary record confirms other records left at much later dates by his descendants.
They can provide evidence of death in times and places where vital records are scarce:
In 1823, Oliver Lee sold part of lot 29 in the Town of Warsaw, Genesee County, New York, to Matthew Hoffman. It was described as "beginning at a stake in the north line of Land owned by Chauncey L. Sheldon..." Nine years later, when Hoffman sold the same land to Isaac C. Bronson, it was described as "beginning at a stake in the north line of land owned by the late Chauncey L. Sheldon deceased..."
In this case, the deeds' information can be confirmed. Dr. Chauncey L. Sheldon has a beautiful and well-preserved 1828 gravestone in the Warsaw Pioneer Cemetery. It's documented and imaged on Find A Grave -- along with other unsourced material that does not appear on the stone. Since 1841 the graveyard has been in Wyoming County, New York, but when Chauncey died it wasn't.
Confirmation doesn't mean the deeds are unnecessary. No important genealogical conclusion should rest on a single piece of information, any more than a chair should have only one leg.
Harold Henderson, "Two Simple Things Deeds Can Do," Midwestern Microhistory: A Genealogy Blog, posted 11 July 2012 (http://midwesternmicrohistory.blogspot.com : accessed [access date]). [Please feel free to link to the specific post if you prefer.]